-
Deep learning of contagion dynamics on complex networks
Authors:
Charles Murphy,
Edward Laurence,
Antoine Allard
Abstract:
Forecasting the evolution of contagion dynamics is still an open problem to which mechanistic models only offer a partial answer. To remain mathematically or computationally tractable, these models must rely on simplifying assumptions, thereby limiting the quantitative accuracy of their predictions and the complexity of the dynamics they can model. Here, we propose a complementary approach based o…
▽ More
Forecasting the evolution of contagion dynamics is still an open problem to which mechanistic models only offer a partial answer. To remain mathematically or computationally tractable, these models must rely on simplifying assumptions, thereby limiting the quantitative accuracy of their predictions and the complexity of the dynamics they can model. Here, we propose a complementary approach based on deep learning where the effective local mechanisms governing a dynamic on a network are learned from time series data. Our graph neural network architecture makes very few assumptions about the dynamics, and we demonstrate its accuracy using different contagion dynamics of increasing complexity. By allowing simulations on arbitrary network structures, our approach makes it possible to explore the properties of the learned dynamics beyond the training data. Finally, we illustrate the applicability of our approach using real data of the COVID-19 outbreak in Spain. Our results demonstrate how deep learning offers a new and complementary perspective to build effective models of contagion dynamics on networks.
△ Less
Submitted 23 June, 2021; v1 submitted 9 June, 2020;
originally announced June 2020.
-
Detecting structural perturbations from time series with deep learning
Authors:
Edward Laurence,
Charles Murphy,
Guillaume St-Onge,
Xavier Roy-Pomerleau,
Vincent Thibeault
Abstract:
Small disturbances can trigger functional breakdowns in complex systems. A challenging task is to infer the structural cause of a disturbance in a networked system, soon enough to prevent a catastrophe. We present a graph neural network approach, borrowed from the deep learning paradigm, to infer structural perturbations from functional time series. We show our data-driven approach outperforms typ…
▽ More
Small disturbances can trigger functional breakdowns in complex systems. A challenging task is to infer the structural cause of a disturbance in a networked system, soon enough to prevent a catastrophe. We present a graph neural network approach, borrowed from the deep learning paradigm, to infer structural perturbations from functional time series. We show our data-driven approach outperforms typical reconstruction methods while meeting the accuracy of Bayesian inference. We validate the versatility and performance of our approach with epidemic spreading, population dynamics, and neural dynamics, on various network structures: random networks, scale-free networks, 25 real food-web systems, and the C. Elegans connectome. Moreover, we report that our approach is robust to data corruption. This work uncovers a practical avenue to study the resilience of real-world complex systems.
△ Less
Submitted 9 June, 2020;
originally announced June 2020.
-
Spectral dimension reduction of complex dynamical networks
Authors:
Edward Laurence,
Nicolas Doyon,
Louis J Dubé,
Patrick Desrosiers
Abstract:
Dynamical networks are powerful tools for modeling a broad range of complex systems, including financial markets, brains, and ecosystems. They encode how the basic elements (nodes) of these systems interact altogether (via links) and evolve (nodes' dynamics). Despite substantial progress, little is known about why some subtle changes in the network structure, at the so-called critical points, can…
▽ More
Dynamical networks are powerful tools for modeling a broad range of complex systems, including financial markets, brains, and ecosystems. They encode how the basic elements (nodes) of these systems interact altogether (via links) and evolve (nodes' dynamics). Despite substantial progress, little is known about why some subtle changes in the network structure, at the so-called critical points, can provoke drastic shifts in its dynamics. We tackle this challenging problem by introducing a method that reduces any network to a simplified low-dimensional version. It can then be used to describe the collective dynamics of the original system. This dimension reduction method relies on spectral graph theory and, more specifically, on the dominant eigenvalues and eigenvectors of the network adjacency matrix. Contrary to previous approaches, our method is able to predict the multiple activation of modular networks as well as the critical points of random networks with arbitrary degree distributions. Our results are of both fundamental and practical interest, as they offer a novel framework to relate the structure of networks to their dynamics and to study the resilience of complex systems.
△ Less
Submitted 12 February, 2019; v1 submitted 21 September, 2018;
originally announced September 2018.
-
Phase transition in the recoverability of network history
Authors:
Jean-Gabriel Young,
Guillaume St-Onge,
Edward Laurence,
Charles Murphy,
Laurent Hébert-Dufresne,
Patrick Desrosiers
Abstract:
Network growth processes can be understood as generative models of the structure and history of complex networks. This point of view naturally leads to the problem of network archaeology: reconstructing all the past states of a network from its structure---a difficult permutation inference problem. In this paper, we introduce a Bayesian formulation of network archaeology, with a generalization of…
▽ More
Network growth processes can be understood as generative models of the structure and history of complex networks. This point of view naturally leads to the problem of network archaeology: reconstructing all the past states of a network from its structure---a difficult permutation inference problem. In this paper, we introduce a Bayesian formulation of network archaeology, with a generalization of preferential attachment as our generative mechanism. We develop a sequential Monte Carlo algorithm to evaluate the posterior averages of this model, as well as an efficient heuristic that uncovers a history well correlated with the true one, in polynomial time. We use these methods to identify and characterize a phase transition in the quality of the reconstructed history, when they are applied to artificial networks generated by the model itself. Despite the existence of a no-recovery phase, we find that nontrivial inference is possible in a large portion of the parameter space as well as on empirical data.
△ Less
Submitted 20 December, 2019; v1 submitted 24 March, 2018;
originally announced March 2018.
-
Exact analytical solution of irreversible binary dynamics on networks
Authors:
Edward Laurence,
Jean-Gabriel Young,
Sergey Melnik,
Louis J. Dubé
Abstract:
In binary cascade dynamics, the nodes of a graph are in one of two possible states (inactive, active), and nodes in the inactive state make an irreversible transition to the active state, as soon as their precursors satisfy a predetermined condition. We introduce a set of recursive equations to compute the probability of reaching any final state, given an initial state, and a specification of the…
▽ More
In binary cascade dynamics, the nodes of a graph are in one of two possible states (inactive, active), and nodes in the inactive state make an irreversible transition to the active state, as soon as their precursors satisfy a predetermined condition. We introduce a set of recursive equations to compute the probability of reaching any final state, given an initial state, and a specification of the transition probability function of each node. Because the naive recursive approach for solving these equations takes factorial time in the number of nodes, we also introduce an accelerated algorithm, built around a breath-first search procedure. This algorithm solves the equations as efficiently as possible, in exponential time.
△ Less
Submitted 13 March, 2018; v1 submitted 7 November, 2017;
originally announced November 2017.
-
Geometric evolution of complex networks
Authors:
Charles Murphy,
Antoine Allard,
Edward Laurence,
Guillaume St-Onge,
Louis J. Dubé
Abstract:
We present a general class of geometric network growth mechanisms by homogeneous attachment in which the links created at a given time $t$ are distributed homogeneously between a new node and the exising nodes selected uniformly. This is achieved by creating links between nodes uniformly distributed in a homogeneous metric space according to a Fermi-Dirac connection probability with inverse temper…
▽ More
We present a general class of geometric network growth mechanisms by homogeneous attachment in which the links created at a given time $t$ are distributed homogeneously between a new node and the exising nodes selected uniformly. This is achieved by creating links between nodes uniformly distributed in a homogeneous metric space according to a Fermi-Dirac connection probability with inverse temperature $β$ and general time-dependent chemical potential $μ(t)$. The chemical potential limits the spatial extent of newly created links. Using a hidden variable framework, we obtain an analytical expression for the degree sequence and show that $μ(t)$ can be fixed to yield any given degree distributions, including a scale-free degree distribution. Additionally, we find that depending on the order in which nodes appear in the network---its $\textit{history}$---the degree-degree correlation can be tuned to be assortative or disassortative. The effect of the geometry on the structure is investigated through the average clustering coefficient $\langle c \rangle$. In the thermodynamic limit, we identify a phase transition between a random regime where $\langle c \rangle \rightarrow 0$ when $β< β_\mathrm{c}$ and a geometric regime where $\langle c \rangle > 0$ when $β> β_\mathrm{c}$.
△ Less
Submitted 4 October, 2017;
originally announced October 2017.
-
Phase transition of the susceptible-infected-susceptible dynamics on time-varying configuration model networks
Authors:
Guillaume St-Onge,
Jean-Gabriel Young,
Edward Laurence,
Charles Murphy,
Louis J. Dubé
Abstract:
We present a degree-based theoretical framework to study the susceptible-infected-susceptible (SIS) dynamics on time-varying (rewired) configuration model networks. Using this framework, we provide a detailed analysis of the stationary state that covers, for a given structure, every dynamic regimes easily tuned by the rewiring rate. This analysis is suitable for the characterization of the phase t…
▽ More
We present a degree-based theoretical framework to study the susceptible-infected-susceptible (SIS) dynamics on time-varying (rewired) configuration model networks. Using this framework, we provide a detailed analysis of the stationary state that covers, for a given structure, every dynamic regimes easily tuned by the rewiring rate. This analysis is suitable for the characterization of the phase transition and leads to three main contributions. (i) We obtain a self-consistent expression for the absorbing-state threshold, able to capture both collective and hub activation. (ii) We recover the predictions of a number of existing approaches as limiting cases of our analysis, providing thereby a unifying point of view for the SIS dynamics on random networks. (iii) We reinterpret the concept of hub-dominated phase transition. Within our framework, it appears as a heterogeneous critical phenomenon : observables for different degree classes have a different scaling with the infection rate. This leads to the successive activation of the degree classes beyond the epidemic threshold.
△ Less
Submitted 12 April, 2018; v1 submitted 26 September, 2017;
originally announced September 2017.
-
Susceptible-infected-susceptible dynamics on the rewired configuration model
Authors:
Guillaume St-Onge,
Jean-Gabriel Young,
Edward Laurence,
Charles Murphy,
Louis J. Dubé
Abstract:
We investigate the susceptible-infected-susceptible dynamics on configuration model networks. In an effort for the unification of current approaches, we consider a network whose edges are constantly being rearranged, with a tunable rewiring rate $ω$. We perform a detailed stationary state analysis of the process, leading to a closed form expression of the absorbing-state threshold for an arbitrary…
▽ More
We investigate the susceptible-infected-susceptible dynamics on configuration model networks. In an effort for the unification of current approaches, we consider a network whose edges are constantly being rearranged, with a tunable rewiring rate $ω$. We perform a detailed stationary state analysis of the process, leading to a closed form expression of the absorbing-state threshold for an arbitrary rewiring rate. In both extreme regimes (annealed and quasi-static), we recover and further improve the results of current approaches, as well as providing a natural interpolation for the intermediate regimes. For any finite $ω$, our analysis predicts a vanishing threshold when the maximal degree $k_\mathrm{max} \to \infty$, a generalization of the result obtained with quenched mean-field theory for static networks.
△ Less
Submitted 6 January, 2017;
originally announced January 2017.
-
Finite size analysis of the detectability limit of the stochastic block model
Authors:
Jean-Gabriel Young,
Patrick Desrosiers,
Laurent Hébert-Dufresne,
Edward Laurence,
Louis J. Dubé
Abstract:
It has been shown in recent years that the stochastic block model (SBM) is sometimes undetectable in the sparse limit, i.e., that no algorithm can identify a partition correlated with the partition used to generate an instance, if the instance is sparse enough and infinitely large. In this contribution, we treat the finite case explicitly, using arguments drawn from information theory and statisti…
▽ More
It has been shown in recent years that the stochastic block model (SBM) is sometimes undetectable in the sparse limit, i.e., that no algorithm can identify a partition correlated with the partition used to generate an instance, if the instance is sparse enough and infinitely large. In this contribution, we treat the finite case explicitly, using arguments drawn from information theory and statistics. We give a necessary condition for finite-size detectability in the general SBM. We then distinguish the concept of average detectability from the concept of instance-by-instance detectability and give explicit formulas for both definitions. Using these formulas, we prove that there exist large equivalence classes of parameters, where widely different network ensembles are equally detectable with respect to our definitions of detectability. In an extensive case study, we investigate the finite-size detectability of a simplified variant of the SBM, which encompasses a number of important models as special cases. These models include the symmetric SBM, the planted coloring model, and more exotic SBMs not previously studied. We conclude with three appendices, where we study the interplay of noise and detectability, establish a connection between our information-theoretic approach and random matrix theory, and provide proofs of some of the more technical results.
△ Less
Submitted 27 June, 2017; v1 submitted 31 December, 2016;
originally announced January 2017.
-
Complex networks as an emerging property of hierarchical preferential attachment
Authors:
Laurent Hébert-Dufresne,
Edward Laurence,
Antoine Allard,
Jean-Gabriel Young,
Louis J. Dubé
Abstract:
Real complex systems are not rigidly structured; no clear rules or blueprints exist for their construction. Yet, amidst their apparent randomness, complex structural properties universally emerge. We propose that an important class of complex systems can be modeled as an organization of many embedded levels (potentially infinite in number), all of them following the same universal growth principle…
▽ More
Real complex systems are not rigidly structured; no clear rules or blueprints exist for their construction. Yet, amidst their apparent randomness, complex structural properties universally emerge. We propose that an important class of complex systems can be modeled as an organization of many embedded levels (potentially infinite in number), all of them following the same universal growth principle known as preferential attachment. We give examples of such hierarchy in real systems, for instance in the pyramid of production entities of the film industry. More importantly, we show how real complex networks can be interpreted as a projection of our model, from which their scale independence, their clustering, their hierarchy, their fractality and their navigability naturally emerge. Our results suggest that complex networks, viewed as growing systems, can be quite simple, and that the apparent complexity of their structure is largely a reflection of their unobserved hierarchical nature.
△ Less
Submitted 10 December, 2015; v1 submitted 30 November, 2013;
originally announced December 2013.
